Summary
Peter Diamandis interviews Ray Kurzweil (Google Principal AI Visionary, co-founder of Singularity University) on AGI timeline (2029, “conservative”), the Singularity (2045), human-AI merger via brain-computer interface in the 2030s, longevity escape velocity (~2032), job displacement, and the law of accelerating returns. Includes the origin story of how Diamandis introduced Kurzweil to Larry Page, leading to Google acquiring Kurzweil’s company Patterns Inc. Kurzweil maintains his 1999 prediction of AGI by 2029 and rejects the notion of a hard takeoff intelligence explosion, instead predicting exponential but continuous growth.
Key Segments
- [00:04] Origin story: Diamandis introduced Kurzweil to Larry Page; Page said “I don’t want to invest, I’ll buy the company” (Patterns Inc)
- [00:07] AGI by 2029 stands firm; Singularity 2045 because merger with AI takes longer than external AGI
- [00:09] Singularity = humans merging with AI via nanorobots, not separate AI surpassing us
- [00:11] No hard takeoff intelligence explosion expected; exponential growth but not discontinuous
- [00:13] Elon called Kurzweil “conservative” on AGI timeline; 2029 prediction made in 1999 (30 years out)
- [00:15] AI risk: 80/20 assessment; problems will exist but tools to address them will also expand
- [00:20] Law of accelerating returns chart: 75 quadrillion-fold increase in computation per constant dollar since 1939
- [00:22] Kurzweil advocates renaming LLMs to “Large Event Models” — they apply beyond language
- [00:27] Job displacement: 80% worked in food production 200 years ago, now 2%; new job types will emerge
- [00:30] UBI coming in 2030s; humans need purpose, will “uplevel” aspirations with greater intelligence
- [00:36] Nanotechnology: Drexlerian assemblers not needed for 2030s brain-cloud interface; only need neocortex top-of-funnel connection
- [00:41] Nobel Prize prediction: almost every future Nobel in chemistry, physics, medicine will be AI-enabled
- [00:43] Longevity escape velocity: by ~2032, each year of scientific progress adds back a full year of life
Notable Claims
- Kurzweil’s prediction accuracy rate: 86% (with 12-24 month leniency)
- AGI by 2029 (unchanged since 1999 prediction)
- 75 quadrillion-fold increase in computation per constant dollar (1939 to 2024, Nvidia B200)
- Longevity escape velocity by ~2032: currently gaining back ~4 months per year from research
- In 1999 Stanford conference, 80% of attendees (including Hinton) predicted AGI in 100+ years
- Kurzweil’s title at Google: “Principal Research AI Visionary”
- LLMs should be called “Large Event Models” (Kurzweil’s proposal)
Guests
- Ray Kurzweil — Google Principal AI Visionary, inventor, futurist, author of The Singularity Is Nearer
Assessment
Essential Kurzweil interview — the definitive statement of his current positions on AGI, Singularity, and longevity escape velocity as of late 2024. The Larry Page acquisition story (Patterns Inc) is a rare inside-baseball anecdote about how Kurzweil ended up at Google. His distinction between AGI (2029, external) and Singularity (2045, merged) is the clearest I’ve seen him articulate it — the extra 16 years are for brain-computer interface maturation, not AI capability. His rejection of hard takeoff intelligence explosion in favor of continuous exponential growth is a meaningful disagreement with the Bostrom/Yudkowsky camp. The “Large Event Models” renaming proposal is clever but unlikely to catch on. The longevity escape velocity prediction (2032, gaining a full year per year) is his most testable near-term claim. Sponsor contamination heavy (Abundance 360 pitch, Fountain Life, Levels, OneSkin) but analytical content is high quality.